📢 Announcing the upcoming GA of UiPath GenAI Activities

News Product News
, , , ,
1

We’re very excited to announce the imminent GA release of UiPath GenAI Activities.

This collection of activities available as a first-party Integration Service Connector allows you to interact directly with popular large language models from various vendors internally managed by UiPath via the AI Trust Layer. No subscription to 3rd party Cloud services is necessary and UiPath takes the guess work out of prompting by offering quality-controlled, common RPA tasks in the form of easy-to-use Studio and Studio Web activities such as:

  • Summarization
  • Translation
  • Language Detection
  • Email Generation
  • Rewrite (content)

Other activities include PII Detection which allow users to review a string of text and identify and redact specific types of PII and PHI.

While we plan on continuing to expand on these curated activities, we are also offering a custom prompt activity called Content Generation. This activity allows you to pass arguments, variables and use the expressions to build resilient, dynamic prompt-templates that are ready to tackle any use case.

image

We’re also excited to announce the Public Preview of Context Grounding which will allow you to ground your custom prompts in specific context either directly uploading a file or referencing an orchestrator bucket/file repository. This will be exposed as an input option in the Content Generation activity. More details to come in a future forum thread.

As noted above, all of these new GenAI offerings take advantage of the governance and auditability offered by the AI Trust Layer.

The team looks forward to following up with other activities such as:

  • Object Detection (images)
  • Categorization
  • Named Entity Recognition
  • Image Classification
  • and others…(happy to hear feedback :slightly_smiling_face:)

The GA date is set for the 23.7 release. Context Grounding will be available at that time in public preview. Please note that until the GA, these activities are free to use. Once transitioned to GA, AI Units will be required to use them in processes.

The team looks forward to hearing about all the cool use cases that can be accomplished! Please give them a shot and let us know your feedback (check docs for how to access while in Public Preview).

5 Likes
2

Regarding the additional activities that UiPath could add based on image recognition are as follows:

  1. Face recognition, full or partial (hoodie), side, back
  2. Height, girth.
  3. Activity recognition
  4. Pattern recognition of the activity recognized (signature of the body language)
  5. Voice signature based on audio.

I’m happy to elaborate more if UiPath has questions.

1 Like
3

Awesome suggestions. We’re hoping to have at least one vision model available as part of the GA. This would allow you to accomplish these kinds of use-cases by crafting a custom prompt.

In the future, we’ll certainly consider these for quality-tested, ‘curated’ activities!

1 Like